Introduction

Linguistic utterances that ascribe moral status to an entity are typically formulated in the form of a descriptive statement that is meant to refer to truths about the entity in question. For example, when it is said that humans are the bearers of fundamental human rights or that non-human animals deserve moral consideration if they can suffer, the language user intends to communicate something about humans and animals that can, in principle, be evaluated as true or false. In arguments concerning moral status, this link to truth(s) about the entity is established by referring to an intrinsic property (or set of properties) of the entity in question. For example, Singer’s (1995) famous argument for animal rights (8) says that animals deserve moral consideration if they have sentience, defined as the capacity for suffering and enjoyment. This is meant to convey a truth about a particular category of animals, which is argued to translate into a truth about their moral status. In other words, such utterances about moral status are based on moral realism: the position that moral sentences are propositions referring to objective features of the world, features independent on human subjects.

In addition, the moral status in question is meant to apply universally, that is, for all entities that belong to this category, regardless of the situation and context. For example, it is said that all animals that are sentient must be given moral consideration, regardless whether they are animals raised for food or animals held as pets. Moral status is supposed to rely on context-independent truths about the morally relevant capacities of the entity. To recall Plato’s famous distinction between doxa and episteme in the Republic and his condemnation of rhetoric in Gorgias: according to this view, moral status concerns truth and true knowledge, not opinion or rhetoric.

Furthermore, moral status is also understood to be independent from the subject that ascribes moral status. The ascription is seen as a description; strictly speaking, it is not even an ascription (since that suggests that subjects somehow “give” it to the receiver of moral status) but a statement about a truth and a representation of a truth. It is supposed to be a truth independent of the activity and subjectivity of human ascribing. It is meant to refer to an objective state of the world. For example, one may hold that a cat has a moral status independently of us humans ascribing that status. To say that it has a moral status is to describe or represent a truth about the world. The “ascription” itself does not influence the moral status. It is not really an ascription but an acknowledgment of what is already there.

However, across different traditions in philosophy of language, it has been argued that there is also a different kind of language use: the performative use of language. The general idea is that language is not necessarily used to say something about truth and the world, but that it can also ‘do’ things (to use Austin’s (1962) famous phrase) and make others do things. As I will note below, this view has a lineage that goes back to Nietzsche and Wittgenstein but in contemporary philosophy has been most influentially formulated by Austin, Searle, and Butler; it is also often combined with Foucault’s view of power. A representational epistemology focused on truth is replaced by a social and political one, and often (but not necessarily) realist views are replaced by constructivist views. Evaluation is then not about truth but about who does what to whom, via language.

In contemporary philosophy, performative views are usually applied to humans – if at all. For example, Butler (1988) has argued that gender and identity are performatively constructed and more recently Mills (2012) has argued for seeing personhood as performative in the context of bioethics discussions, in particular abortion. Yet in spite of decades long discussions about the moral status of animals, for example, performance thinking has not yet been widely applied to non-humans. Moreover, typically arguments about performative language use are not applied to moral status or even morality in general. While many would be willing to grant that performative language use is used in politics (traditionally the domain of rhetoric and doxa as public opinion) and would follow Searle in acknowledging the performative construction of social reality in the sense that our language use has social effects, moral status is a domain that tends to be seen as somehow exempt from performativity, perhaps because it is feared that otherwise morality may be contaminated by moral relativism. Moral realists accept that our speech acts have practical effects, but maintain that when it comes to deciding moral status the basis is still the intrinsic properties of the entity in question. For example, moral realists may accept that speech acts about robot rights have political effects, but maintain that this does not entail that their moral status is constructed by those speech acts.

Drawing on Austin, Searle, and Butler and further developing my (Coeckelbergh, 2012) view that language is a condition of possibility for moral status ascription, this paper explores what it would mean to say that moral status itself is performatively constructed. It distinguishes between a strong and a weak claim about the performativity of moral status utterances. Then it discusses the implications for thinking about the moral status of humans but also nonhumans such as non-human animals and anthropomorph robots or artificial intelligence (AI), responding to the current discussion about the moral status of robots and AI. This aims at helping to highlight and understand the performative dimension of utterances about moral status made in those domains, for example utterances made in public on digital social media – today an important forum for such discussions.

Performative approaches to language use and its application to moral status ascription

Nietzsche already held the view that language always mediates what we say about the world; there is no access to truth independent of language. Language is the condition of possibility of our thinking. Language is rhetoric and there is no truth outside social convention (Nietzsche, 1979; see also Pakhurst, 2019). The later Wittgenstein (2009), in turn, argued that the meaning of language must always be understood as related to the context of use, in particular what he called games and a form of life. Both views prepared a more social-epistemological understanding of language and its relation to the world, which later was taken up by postmodern thinkers. For example, Foucault further built on Nietzsche and Lyotard and Derrida referred to Wittgenstein. But the most explicit and influential formulations of a performative view of language are to be found in Austin and later in Searle and Butler.

In his speech act theory, Austin identified a type of linguistic utterances that do not describe reality but do something. In How to Do Things With Words (1962), he showed that speech acts have illocutionary force and perlocutionary effect: speakers who use performative utterances do something with words and want their words to have a particular effect. These utterings thus are, or are part of, doing something (1962: 5). A typical example is a marriage ceremony, in which language is used not to describe a reality but to create one: ‘I. . take this woman to be my lawful wedded wife’ does not represent reality but creates a reality: the marriage. These words are an act themselves. They are performative. Austin calls them speech acts. Illocutionary and perlocutionary acts are classes of speech acts.

Searle, influenced by Austin, further developed speech act theory and, in his later work, argued that social ontology is based on the performative use words, on speech acts. In particular, it is based on what he called ‘declarations’ (Searle, 1995). What does that mean? In contrast to physical facts, Searle argues, social facts are observer dependent and are created by language users, which constitute social reality by collectively accepting to give a certain status to a person or object. The basis is collective intentionality and agreement. For example, money can perform its function and be a social fact only because we give it that function by using language. This status ascription is done by means of a specific kind of language: we declare that a particular person or object has a specific status in a particular context. Searle (2006) calls this a ‘status function’, which has the form of ‘X counts as Y in context C’ (18). For example, we declare that this piece of paper counts as money in the context of trade. Again, this use of language is performative: it does not describe or represent reality but constructs at least part of reality: social reality.

Butler refers to Searle’s speech act theory in her work on performativity and gender (identity), next to the phenomenology of Husserl, Merleau-Ponty, and Mead (Butler, 1988: 519). She argues that gender is not a stable identity but is constituted by acts: gender acts. Those performances are not expressive of a pre-existing gender, but create and constitute the gender reality. Here performativity goes beyond language use. Gender is not only constituted by linguistic performances but by all kinds of performances: by language, but also by gesture for example. This involves exercising power. Moreover, for this kind of gender construction to succeed, there needs to be iteration: gender reality needs to be performed over and over again. And in contrast to Searle’s view, the social agent herself is not entirely independent from the performance; instead, she is partly the outcome of performances and is performed.

What do these performative theories do when we apply them to moral status ascription? What does it mean to say that moral status is performatively constructed?

First, it could mean that moral status is constituted by speech acts in the sense that those speech acts are intended to have effects on people to take action and protect the entities in question. Based on Austin’s speech act theory, one could highlight the illocutionary and perlocutionary dimension of speech acts that ascribe moral status. In contrast to the usual view that moral status utterances are a description of truths in the world (in particular truths about entities based on truths about their properties), utterances concerning moral status are then at least also, if not only, speech acts that create the reality of the moral status. For example, when we say that refugees have human rights, then one could argue that this is not only and not so much meant as a truth about refugees and a description of reality; it is also doing something (granting refugees these rights, protecting them) and intends to have an effect on others, which are supposed to recognize these rights and protect these people. Thus, it is a performative use of language, in particular a performative use of language to ascribe moral status.

Furthermore, drawing on Searle’s theory of social ontology one could see moral status as a social fact that is created by status declarations. Let me call these moral status declarations. Via the performative use of language, in the form of declarations, we create a particular moral status in a specific context. For example, written declarations of human rights and the related ceremonies can be seen as constituting human rights as a social fact. We declare all humans to be the bearers of these rights, and with these speech acts these rights come into existence and become a social reality. Once they are declared, we can appeal to them when needed. Such moral status declarations are thus done in a particular context and are also applied in specific situations, for example when human rights are violated.

With Butler, we can then expand the range of acts that count as performative with regard to moral status. Not only speech acts, but also other acts create moral status as a social and performative reality. For example, to establish human rights, it is important that one also does other acts that confirm that all humans have these rights. These performances also have to be iterated. Human rights have to be performed over and over again, in the sense that there need to be always again declarative statements about the human rights (and other acts) concerning particular people in particular contexts. Such a moral status is then not something that “is” already in the world but that needs to be performed repeatedly in order for it to be(come) reality. With our utterances concerning moral status, we do not only say what is true and real, but also exercise power and try to change something in the world.

Thus, according to this performative view, moral status ascription is not only and not primarily a matter of saying something about truth, reality, and metaphysics; rather, understood in a performative sense, moral status ascription is deeply rhetorical, social, and political. It is an act and a performance, which does things (granting moral status) and tries to have others do things (have others respect that moral status), and thereby change social reality. Moral status ascription, understood performatively, is a way by which we constitute moral status as a social reality.

Now moral realists could accept that moral status is constructed by speech acts in the sense that these speech acts are meant to have an effect on the world. However, they would resist interpreting that this performativity touches the basis of moral status. They would maintain, for example, that human rights are based on, and justified by, the reality of properties that humans have. The view I am interested to explore in this paper, however, questions if there is a moral reality that can be entirely separated from the way we know and socially construct it through language and otherwise. It sees moral status itself as linguistically-performatively constructed and as political. In that sense it could be called a “deep” or “strong” performative view.

Such a view would be in line with the relational turn Coeckelbergh (2012) and Gunkel (2018) have developed in the area of thinking about the moral status of robots. According to such relational approaches, the moral status of robots does not depend on intrinsic properties of the robot but on the ways we talk about robots and decide about their status. This could be put in performative language: it depends on our linguistic performances. While this can also be seen as a way of understanding how moral status works, as a normative view it invites a well-known objection: relativism. Let me discuss this objection in order to further develop and support the deep performative view of moral status that understands this status as itself performatively constructed.

Discussion of objections and strong and weak interpretations of the deep performativity of moral status ascription

A common objection to these kinds of views is the charge of moral relativism. For example, in the discussion about robot rights and in response to Danahar, who has interpreted relational views in terms of ethical behaviorism, Müller (2021) has argued that if moral status depends on how we feel towards and entity or how we decide to respond, ‘the relational turn is a relativist account of moral status’, having the implication that anything goes (582). At most, he argues, entities such as robots can have derived moral status (see also what I have called ‘indirect’ moral status in recent work, e.g. Coeckelbergh, 2021). Similarly, one could object to the deep performative view of moral status that if moral status is just a matter of linguistic performance, rather than the intrinsic properties of an entity, we invite moral relativism and anything goes. We want to avoid this, for example with regard to human rights: we want to make sure that their basis remains untouched by relativism.

A response then needs to ask: what kind of moral relativism do you mean? Performative views of moral status are certainly problematic if we presume that moral status entails a set of subject-independent moral facts that are not created by humans or agreed upon by humans in any sense. A deep performative approach rejects this view of moral status and is indeed relativistic in the sense that it sees moral status as dependent on linguistic and other forms of acts and performances and on agreement – it is linked to what Searle called collective intentionality. It thus rejects the moral realist position and offers a social account of moral status.

However, such a deep performative approach does not mean that “anything goes”, an accusation that is also often voiced against postmodernism. Once a performative act (or series of acts) establishes moral status of a particular entity as a social fact, this is a social fact, that is, it is a social reality that humans in that particular context and community have to reckon with and have to respect. For example, once universal human rights are performatively established, it is a social fact and cannot be easily ignored or disrespected. As Durkheim and Weber have argued, social facts transcend the psychology of individual agents, create normative expectations concerning other agents’ behavior, and exercise social control. Social facts such as marriage, currency, and religion establish constraints and they take on an existence of their own. And this is not necessarily problematic. In the case of human rights, for example, we want human rights to be a social fact and provide constraints. In other words, we do not want that human rights are just up to the individual to decide whether to apply them or not. We agree on that. Relativism in that sense is thus avoided. Of course the social agreement will have constraints that are due to the way a particular community and society is constituted: it will have its own moral values and norms and its own moral horizon. In this sense, performative approaches may involve relativism. Yet while such values and norms are still relative to the particular social and cultural context, within that context the ascription of moral status, understood in terms of performativity, not everything is possible. Moral status is relative to the particular society but is not just a matter of what individuals believe, think, and want. Performative views may thus incur a form of cultural moral relativism but not an individualistic form of relativism.

Moreover, linguistic utterances still need to make sense, and this also puts some constraints on what can be said about an entity in terms of its moral status, since utterances that don’t make much sense in a particular social context will not have much performative effect. For example, if I would claim today that some humans have no moral status whatsoever, my claim would most likely be ignored, since it does not (or no longer) make sense to claim this, based on the collective knowledge and experience we have today. The linguistic and social aspect of performativity is not only a danger (what the majority of people believe is not necessarily right) but also guards against making such utterances: their performative effect will be rather small.

These responses to the charge of relativism thus bite the bullet: they reject moral realism and endorse moral relativism, but specify that it is a specific form of moral relativism (cultural and descriptive moral relativism), and not “vulgar” moral relativism concerning individual opinions and (unjustified) beliefs that the interlocutor may have had in mind and of which postmodern thinkers are often accused.

Another response is to argue that this cultural moral relativism is only entertained at the descriptive level, and not at the normative level. One may acknowledge that moral status ascriptions, understood in a performative way, vary according to social and cultural context, but refuse to make a similar normative claim, that is, refuse to say that moral status ascriptions should vary culturally. In that case the issue is avoided altogether.

However, one could also question and deconstruct the very distinction between normative and descriptive, in so far as it is linked to moral realism and moral fundamentalism. Pragmatist (meta-)ethics rejects a view of morality in terms of eternal truth independent from human societies, and promotes a situational, social, and pluralistic view of morality. According to this view, moral status does not exist in a separate eternal realm of value as opposed to facts. Moral status is a practical and social problem that we need to try to resolve in a situation (Pappas, 2008; Fesmire, 2003). Performative uses of language and other acts, then, could be seen as ways to deal with situations in which moral status comes up as a social and societal problem. For example, the question concerning the status of a human embryo or the question regarding the status of refugees come up in a particular situation and society. Performative utterances about moral status of embryos or refugees are then ways of coping with this problem in that situation and that particular society.

Does that mean there is no moral truth about moral status? An opponent could align performative views of moral status with a more radical postmodern interpretations of Nietzsche, according to which morality is only a matter rhetoric and social convention. Moral status, according to this view, has nothing to do with truth and is just a matter of agreement and power. What the character of Protagoras says in Plato’s famous dialogue can also be interpreted as such: truth and morality are relative; moral utterances are true or false only relative to a particular social context.

However, few philosophers actually endorse such an extreme view and it seems there is a good way to avoid it. One may rather embrace a Kantian interpretation of Nietzsche, which takes an agnostic position and stresses that there may well be moral truth in a noumenal realm but that this is not (directly) accessible to us and that therefore we have to do with rhetoric and its moral performances. Moral status ascriptions are then performances of moral status that do not penetrate into the realm of truth (to use Nietzsche’s metaphor picked up by Pakhurst (2019).

Furthermore, a Kantian angle may also help to limit the performative view to the descriptive: one may accept the performativity of moral status in so far that moral status is a social fact (this is what Searle’s view implies), but reject the performativity of moral status in so far as this claim moves into the moral domain. In other words, one may accept that, descriptively speaking, moral status is a matter of speech acts and performances, but reject it as a claim about normative moral status. In that way, one “protects” the Kantian moral realm while at the same time allowing for a performative explanation of how we actually go about moral status. This would avoid the charge of moral relativism.

These Kantian responses thus lead us to consider the possibility of a more modest, weaker form of a deep performative view of moral status, a form that limits the scope of the performative view of moral status ascription. I propose to distinguish between two versions that moral status is performative: a strong one and a weak one:

According to the strong view, moral status is performative in a normative sense and thus its performativity goes “all the way” to the moral. It sees moral status in terms of moral status declarations that establish a particular moral status of particular entities as a social fact and there is nothing more to moral status and morality than social facts created by speech acts and other performances. Moral status utterances create and maintain social conventions. But, contrary to what Searle assumed, this is not just a matter of creating social facts as opposed to others. Performativity goes beyond the boundaries of the social narrowly conceived. There is no wall between the moral and the social, between the moral and the performative, between the moral and language. This is the Nietzschean and perhaps also pragmatist position. Moral status is constructed by means of performative language use and other acts, which take place in particular situations and contexts, and are meant to cope with these situations and contexts.

According to the weak view, by contrast, the claim about the performativity of moral status is “deep” in the sense that it reaches to the basis of moral status, but is only meant descriptively: it is about how we actually think and “do” moral status by means of the performative use of language, about how we in that sense “create” or “construct” moral status, but from this does not follow that moral status actually is a matter of performance only; this is a different question and the question is answered negatively. Moral status cannot and should not be reduced to its performative constitution. Its performative nature goes only as far as moral status is a social fact. The deep performative view helps us to understand how that works. But in so far as moral status has a normative character and connects to the realm of morality, performativity does not touch it. Theory about the performativity of morality and moral status is only applicable to the social domain, which must be strictly distinguished from the moral domain proper. Performance, power, and rhetoric belong to the social, not the moral domain strictly speaking.

The latter view seems an easy way out of the charge of relativism and is tempting. However, if we take the social-epistemological power of the mentioned theories seriously, especially their critical and social-constructivist versions, it is not clear why the performative does not “spill over” to the normative and moral realm. If we cannot know the moral realm – and hence moral statuses – independently from human subjects and their linguistic performances and other acts, as the deep view says, then how effective are the conceptual distinctions that are supposed to keep us from going all the way? For example, if I accept that human rights can only be known through language, then why is this only relevant to the anthropological-descriptive and not to the moral-normative? And is not such a claim about moral status itself a particular (meta-)moral performance, which does something and aims at getting others to do something? To use a famous metaphor that is often used in applied ethics: here seems to be a slippery slope towards the strong version.

A proponent of the weak version could try to maintain the walls. For example, she could argue that while it is true that language always mediates our utterances about morality and moral status, performativity is only one type of language use. We also have other types of language use available to us, including language use that says something about truth and reality regardless of the social and regardless of linguistic performances and rhetoric. Language use is not always performative. Therefore, she could argue, one needs to distinguish between, one the one hand, the performative utterances about moral status that take place in the social realm, and on the other hand, those utterances about moral status that constate something (to use Austin’s term), that say something about the world and that can be evaluated as true or false. For example, she could argue, it may well be in that in politics people use the concept of human rights and get that concept to do all kinds of things by means of rhetoric and performative use of language about human rights (for example justify war or protest against war), but whether or not all humans really have those rights is not a matter of performative language use or rhetoric but of moral truth, which is independent of what humans say and do.

Yet maintaining this distinction is not so easy as it might seem: the very claim that “all humans really have those rights” is itself not just constative and at least also performative. Addressing her opponents, the speaker who argues for human rights in a particular situation already does something with words and tries to make them do things. This also means: make them do things in the moral realm: make them act differently, for example in a way that respects human rights. It seems that the performative is rather expansive and pervasive. It is not easy to contain it within the realm of the social and the descriptive as opposed to the moral and the normative. To the horror of true Kantians, performativity thus “contaminates” the moral realm and does not stay within the boundaries of the social-pragmatic and social-descriptive.

Much more can be said about this. However, I now propose to take a step further and take a critical stance to the performative use of the words of this paper: Imagine that we take on board this deep and strong view, or at least offer it for further discussion, what goals do we want to reach with this kind of philosophical performance about moral status and indeed what do I want to do with this paper and its performative utterances? Or to ask it in a way that is less critically potent and more naïve about the performative aspects of philosophical discourse and that is more common: what are the practical implications of a performative view of moral status ascription?

In order to pursue this direction, let me first zoom in on a particular discussion concerning the moral status of non-humans: that about the moral status of robots and AI.

Implications for thinking about the moral status of humans and nonhumans; Philosophers and scientists as performers

One of my rationales for writing this paper is to critically comment on discussions about the moral status of non-humans, in particular the discussion about the moral status of robots and AI. Some authors in this discussion have argued that we should at least consider giving some robots moral status. For example, Gunkel (2018) and Gellers (2021) have asked the question regarding robot rights. Typically, objections to such proposals and philosophical explorations take what I (Coeckelbergh, 2012) called a ‘properties approach’ to moral status ascription: they refuse to give robots rights – for example, because, as Bryson (2010) has argued, they are not persons – or, in the case of Birhane and Van Dijk (2020), even deny that robots are the kind of beings that could be granted or denied rights since, in contrast to humans, robots lack a lived embodied experience embedded in social practices. Often such debates have a public character, for example on social media, where some fiercely resist the very idea of giving moral status to robots. For example, responding to a paper by Parviainen and Coeckelbergh (2020), which offers a critical perspective on robot rights, Bryson has aggressively criticized the authors on Twitter – generally known as a very political digital social medium – for even using the term “citizenship”, rather than “honorary citizenship”, in conjunction with the mentioning of the Sophia robot.Footnote 1 The background of such reactions is moral and scientific realism: we know with certainty that robots are not the kind of beings that deserve moral status; to deny this is to make an error about the truth and reality of the world. The charge is thus that authors such as Gunkel, Coeckelbergh, Gellers, by even suggesting that robot rights are an appropriate topic for discussion, are mistaken.

How can using the performative view of moral status ascription developed here help with intervening in this discussion? For a start, one can better understand why moral and scientific realists such as Bryson get so upset about the idea to grant robots moral status. Paradoxically, one of the reasons is that what they do with words is not a mere description of reality but at least also a performance. If we see ascriptions of moral status (and indeed even representations of ascriptions of moral status) as merely saying something about the world without having any performative meaning, then it is difficult to understand why a discussion about the moral status of a robot (more precisely: political status) can get so heated. If it was really a matter of getting it right, of mistakes and errors, of truth, then science and theory would help us to find out the truth of the matter. Then having a debate about it does not really make sense. We need to figure out what the truth of the matter is, and according to moral realism this is itself not up to discussion. We should have a scientific discussion about morally relevant properties of the entity in question and correct each other, leaving emotions behind.

However, if we understand such discussions about moral status as performative exercises mediated by language, we can analyze them in terms of performative utterances, and not only in terms of constative ones. We can analyze their rhetoric. We can then ask questions such as: What does this person do by saying this? What does she want to achieve? What did the authors do when they wrote their article, and what effect do they want to have on others? Here potential tensions and even clashes between different interests and goals open up. For example, the initial goal of Parviainen and Coeckelbergh was to describe the debate about the moral and political status of robots as presented in the media and offer an analysis of the politics of robotics as a critical gesture within academia. But the main commentator in question, Bryson, presumably wanted something different: she wanted to stop a particular way of talking about robots and make others talk differently about robots – in academia and elsewhere. She was engaging in a wider kind of politics of robotics and AI: her aim was not so much to criticize the paper as such, but to stage a performative and political gesture about how companies, governments, and other actors should talk about robots. And by doing that, she did not only hope to influence those actors but also incidentally made one the authors aware of his own linguistic performativity, which apparently crossed the boundaries of academia and became itself more political than expected. The point I want to make by taking a performative approach here is not to say, for example, that Coeckelbergh and Gunkel “got it right” whereas Bryson was “performative” and “political”; such a comment would do nothing more than mirror Bryson’s argument and would remain within a moral realist framework. Moreover, it is perfectively legitimate to do a political-performative intervention. Rather, from a performative point of view, it must be recognized that speech acts on both sides were performative and political (next to constative), and that both speech acts have to be analyzed as such in order to fully make sense of that particular debate and indeed to discuss the politics of robotics and AI. (Part of that analysis should also be the role of the very medium, Twitter, in creating an intense and potentially polarized political discussion. However, I will not further develop this here.) Discussing and enhancing awareness about the performative aspects of what philosophers and scientists say about the moral status of robots and AI is important if we want to create, use, and regulate robots in a responsible way. And one outcome of this may well be that we decide to be more careful when using terms such as “rights” or “citizenship” when we talk about robots – as long as we recognize that this is not just a matter of science and technology (doing things with things) but also of politics and rhetoric (doing things with words). And ultimately, we might recognize that science and technology itself are necessarily penetrated by politics: scientists, too, do things with words. What roboticists and AI developers say about (their) robots and AI systems is not politically neutral; this is partly what Parviainen and Coeckelbergh wanted to show and what we need to assume in order to make sense of Bryson’s Twitter intervention.

More generally, the performative view delivers a way of talking about the moral status of humans and non-humans that highlights the political dimension of what we do with words when we talk about moral status. This might be particularly helpful in debates about the moral status of (some) non-humans: whereas most philosophers tend to be in agreement about the moral status of humans, for example about human rights, there is still considerable discussion about non-humans such as non-human animals and robots. Typically, in such cases moral realism is not always performatively successful, since in the cases under discussion it is not clear, scientifically speaking, what morally relevant properties the entities have, and there is no established moral framework that presents the matter as if there are some truths we hold “to be self-evident”, as the U.S. Declaration of Independence puts it. In cases such as the moral status of fish or the moral status of advanced robots and AI, things are not at all that evident and clear. For example, it is not clear whether fish can feel pain. And in the near future it may not be very clear at all whether or not we interact with another human being or an artificial agent when we are on the Internet. In such situations, it becomes both more apparent and more useful to understand that what we say about moral status is itself at least co-creating that moral status and is meant to influence others. For example, there used to be no deontological framework for thinking about the rights of rivers or ecosystems; such a framework had to be constructed by means of speech acts and other performances. Recognizing the role of human subjects and their language then means recognizing our (collective) responsibility as co-creators of moral status reality. Whatever the moral truth about the matter (supposedly existing in some noumenal realm), what status we ascribe by using language in a performative way matters at lot to these non-human entities in question. As a society we better discuss well what kind of words we want to use: what we say about animals, ecosystems, robots, and AI. It is our collective responsibility.

However, even in cases where we already have established moral and political frameworks, such as the framework of human rights, our utterances about moral status have a performative dimension. First, this is definitely true historically speaking: frameworks such as universal human rights have come into existence by means of performative acts, including speech acts. The texts we have today are records of these speech acts and traces of the social and political struggles that helped to performatively constitute this kind of moral status. Consider the speech acts of politicians, for example, but also the speech acts and other acts of those who protested and struggled for their rights or, in the case of non-human animals and ecosystems, had others protest and struggle on their behalf. Second, this is also still relevant today. Keeping in mind Butler’s emphasis on iteration, it is important to see that the frameworks of moral status we have today are never just given and should not be taken as natural but need to be continuously performed in order to be maintained and continue to become real. For example, according to a performative approach to moral status, universal human rights mean nothing if they are not performed through acts – including speech acts – that confirm them in particular situations and towards particular people. The same is true for the moral status of non-human animals. If the relevant speech acts and other acts are not repeated in particular situations (for example situations of abuse of animals), the “moral status” of these particular animals means very little indeed. Recognizing the performative dimension of moral status (and thus at least embracing the weak performativity view) is thus key both for maintaining a particular moral order and for moral and societal change.

Whatever eternal moral truths and realities there may be, this is a huge advantage the performative and pragmatic approach can deliver. Paradoxically, even declaring, in a moral realist way, that a particular category of animals has a particular moral status for moral realist reasons, can contribute to changing the world, and it can do so since it is also performative. To appeal to eternal truths is also a performance with words. And what we say matters, also with regard to moral status and also if we believe that this moral status is somehow given, is somehow part of the world. A weak version of the view explored in this paper could potentially be combined with a version of moral realism, even if that position might not be very stable, given that we do not have direct, unmediated access to the given. But at the very least, and regardless whether one chooses the strong version or not, a performative approach explains can create more awareness about that normative potential of speech acts and other acts. When Singer wrote his Animal Liberation, for example, he framed this argument in a realist way: animals have the capacity for suffering, and therefore they have a moral status. But his utterances – in his book and elsewhere - also helped to change the situation of some animals, the way many readers thought about their food, and so on. The words mattered politically; they were politically performative. Similarly, what we say about robots and AI, as users but also as stakeholders in processes of development and policy, is not only about what robots and AI are but also about what they could and should become. By means of speech acts and other acts, we performatively create their reality and their status. If we “do robots with words,” then their status is our responsibility.

This “create their reality” is important when it comes to highlighting the additional value of the deep performative view: authors such as Bryson might well accept that their utterances have political effects, but deny that they co-create the moral status realities they talk about. The deep performative view, by contrast, goes all the way to recognizing its inherent political and constructed character, including the mediation by language, which enables us to analyze the rhetoric of the debate and to gain a more critical attitude towards our use of language as philosophers. Whereas moral realists claim that what they are doing is merely describing or representing what is right and true, for example by saying that robots are just machines, what they are actually doing is helping to establish a social-linguistic world in which robots get that status and are treated in particular ways rather than other ways.

To conclude, also with regard to moral status can it be said that words do things, and not only in the “social” world: they also at least co-create moral status. With our moral status declarations, we do not only describe what is but also bring the “should”, the normative into being. If this is so (that is, if the deep performative view of moral status gets it right), then we have a significant and important responsibility as users of language and, more generally, as actors and performers within a world that is always already social and mediated by language. Then we are not only the re-cognizers of moral status and value that we find in the world but also its performers. And if we put emphasis on that link between the given and the performed and take an iterative approach, we are not so much unlike those pre-modern people who believed they had to perform and re-perform the moral and metaphysical order of the world.